Skip to main content
Internet Archive's 25th Anniversary Logo

Full text of "Reconstruction of radial temperature profiles of galaxy clusters"

See other formats

A&A manuscript no. 

(will be inserted by hand later ) 

Your thesaurus codes are: 
11.03.1; 12.04.1; 14.25.2 



Reconstruction of radial temperature profiles of galaxy 













Yan-Jie Xue and Xiang-Ping Wu 

Beijing Astronomical Observatory and National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100012, 

Received 28 June, 2000; accepted 27 July, 2000 

Abstract. In this Letter we present the radial temper- 
ature profiles of three X-ray clusters (A119, A2255 and 
A2256) reconstructed from a combination of the X-ray 
surface brightness measurements and the universal density 
profile as the underlying dark matter distribution. Our al- 
gorithm is based on the hydrostatic equilibrium for intra- 
cluster gas and the universality of the total baryon fraction 
within the virial radius. The scaled temperature profiles 
of these three clusters appear to be remarkably similar in 
shape, reflecting the underlying structural regularity, al- 
though they are inconsistent with either isothermality or 
a significant decline with increasing radius. Nevertheless, 
we find a good agreement between our derived tempera- 
ture profiles and the recent analysis of 11 clusters observed 
with BeppoSAX (Irwin & Bregman 2000), which provides 
a useful clue to resolving the temperature profile discrep- 
ancy raised recently in literature. A comparison of our de- 
rived temperature profiles with future spatially-resolved 
spectral measurements may constitute a critical test for 
the standard model of structure formation and the con- 
ventional scenario for dynamical properties of clusters. 

Key words: cosmology: dark matter 
general — X-rays: galaxies 

galaxies: clusters: 

1. Introduction 

The lack of robust constraints on the radial temperature 
profiles of hot gas contained within galaxy clusters is prob- 
ably the major uncertainty in the present determination 
of dynamical properties of clusters, which hinders clusters 
from acting as an ideal laboratory of testing theories of for- 
mation and evolution of structures in the universe includ- 
ing a direct estimate of the cosmic mass density param- 
eter flyi by combining the baryon fraction measurement 
and the Big Bang Nucleosynthesis. Indeed, previous stud- 
ies have arrived at conflicting results regarding the radial 
temperature gradients in clusters. By analyzing 30 clusters 

Send offprint requests to: Y.-J. Xue 
Correspondence to: 

observed with ASCA, Markevitch et al. (1998) claimed a 
significant temperature decline with radius quantified by 
a polytropic index of 1.2-1.3 on the average. However, sub- 
sequent studies have soon raised doubt about the ubiquity 
and steepness of these temperature decline: Irwin, Breg- 
man & Evrard (1999) carried out an analysis of the color 
profiles of the same clusters used by Markevitch et al. 
(1998) but found an essentially flat temperature profile. 
Applying the spectral-imaging deconvolution method to 
a large sample of 106 ASCA clusters, White (2000) has 
showed that 90 percent of the temperature profiles are 
actually consistent with isothermality. Further argument 
against the nonisothermality of intracluster gas has been 
put forward recently by Irwin & Bregman (2000), who 
reported the detection of a flat and even increasing tem- 
perature profile out to ~ 30% of the viral radius for a 
sample of 11 clusters observed with BeppoSAX. 

Theoretically, it deserves an investigation into the pos- 
sibility of deriving the radial temperature profiles of intra- 
cluster gas from the well-motivated physical mechanisms, 
incorporated with the X-ray imaging observations. This 
may provide a valuable clue to resolving the above temper- 
ature profile discrepancy. There are two well-established 
facts on which we can rely today: (l)The gravitational 
potential of a cluster is dominated by the dark matter 
distribution which can be described by the so-called uni- 
versal density profile, as suggested by a number of high- 
resolution simulations (Navarro, Frenk & White 1995 and 
hereafter NFW), although the innermost slope is still un- 
der debate. (2)The azimuthally-averaged X-ray surface 
brightness of a cluster is reliably measurable out to several 
or even ~ 10 times as large as the X-ray core radius, for 
which a good approximation is provided by the conven- 
tional (3 model (Cavaliere & Fusco-Femiano 1976). These 
two facts, along with the hydrostatic equilibrium hypoth- 
esis and a reasonable choice of the boundary conditions, 
permit a unique determination of the gas temperature pro- 
file (Wu & Chiueh 2000). On the other hand, a comparison 
of the theoretically expected temperature profile with the 
result from the X-ray spectroscopic measurement consti- 
tutes a critical test for the validity of the NFW profile and 
the hydrostatic equilibrium in clusters. 

Yan-Jie Xue and Xiang-Ping Wu: Temperature Profiles of Clusters 

In this Letter, we will attempt for the first time to 
derive the temperature profiles of 3 well-defined clusters 
with good X-ray imaging observations extending to rela- 
tively large radii, based on the method developed by Wu 
& Chiueh (2000). Our derived temperature profiles will be 
compared with the recent results of 11 clusters observed 
with BeppoSAX (Irwin & Brcgman 2000). We will exam- 
ine the possible similarity in the gas temperature profiles 
as a result of the underlying structural regularity (e.g. 
Neumann & Arnaud 1999). The implication of our results 
for the reported temperature profile discrepancy will be 
discussed. Throughout the Letter we assume H = 50 km 
s _1 Mpc -1 and a flat cosmological model with J7m = 0.3 
and J7a = 0.7. 

2. Theoretical expectations 

We briefly summarize the mathematical treatment of the 
intracluster gas tracing the underlying dark matter dis- 
tribution of clusters. First, if the X-ray surface brightness 
profile of a cluster can be well approximated by the con- 
ventional (3 model (Cavaliere & Fusco-Femiano 1976) 

/ r 2\ -3/3+1/2 

Sx(r) = So f 1 + ^J , (1) 

this would indicate (Cowie, Hcnriksen & Mushotzky 


3 (r)TVV)=n c0 T 1/4 1 + 

for an optically-thin, thermal brcmsstrahlung emission, 
where n e and T are the electron number density and tem- 
perature, respectively. The central electron number den- 
sity n e o, temperature To and X-ray surface brightness So 
are connected by 

4tt 1 /2 T(3/3) So(l + z) 4 


a{T )p c gT(3f3- 1/2) r c 


where a(T ) = (2 4 e 6 /3m c hc 2 )(27:kT a /3m c c 2 ) 1 / 2 , p c = 
2/(1 + X) with X being the primordial hydrogen mass 
fraction, g w 1.2 is the average Gaunt factor, and z is the 
cluster rcdshift. The total mass in gas within r is simply 

■Mg as (r) = 4ir/j, e m p n c0 




r 2 dr.{4) 

Secondly, if the intracluster gas is in hydrostatic equilib- 
rium with the underlying dark matter distribution, we 



d{n e kT) 



r^ pm p n c 

where p = 0.585 is the average molecular weight. For 
NFW profile 

M DM (r) = 47T/9 s r s 3 

In 1 

r + r s 


Here we have neglected the self-gravity of the gas. Us- 
ing the normalized gas temperature T(r) = T(r)/To 
and the volume-averaged baryon fraction fb{f) = 
M gas (r)/MY)M(r) as the two variables, we obtain the fol- 
lowing two first-order differential equations 




3x 2 

X + a 

a% _ bf- l ' \l + x 2 /a 2 )-^' 2 x 2 

ln(l + x) 


f h x/(l + x) 2 

rn(l + x) -x/(l + x) 


where x = r/r s , a = r c /r Sl b = ^ c n c0 m p /p s and a = 
4irGfj,m p p s r 2 /kTo. The first equation can be straightfor- 
wardly solved with T(0) = 1: 






x ln(l + x) - x/(l + x) 

x 2 (l+x 2 /a 2 ) 2 P 


In order to solve the second equation and determine the 
free parameters a, b and «0j w e use the following boundary 

/b( r vir) — /b,BBN! 



= 0. 



Namely, we demand that the baryon fraction should 
asymptotically match the universal value of /b,BBN at the 
virial radius r v ; r defined by 


(2) M DM (r vir ) = — < ir A c p, 


where A c represents the overdensity of dark matter with 
respect to the average background value p cr ;t , for which we 
take A c = 178£^ 45 (z) and Q. M {z) = n M (l+z)/{l+zn M + 


1]£!a}. We now come to the free parameters 

involved in eqs.(8) and (9). With the X-ray imaging obser- 
vation, we can obtain the best-fit values of /3, r c and So- If, 
on the other hand, the X-ray spectroscopic measurement 
can set a useful constraint on the central temperature To, 
we will be able to derive the central electron density from 
eq.(3). As a result, there are only two free parameters 
in the above equations: p s (or equivalently S c — p s / p C rit) 
and r s . These two parameters can be fixed during the nu- 
merical searches for the solution of eqs.(8) and (9) using 
the boundary conditions eqs.(10) and (11). This will al- 
low us to work out simultaneously the radial profiles of gas 
density and temperature, and fix the dark matter (NFW) 
profile of the cluster characterized by p s and r s . 

3. Application to X-ray clusters 

Since the reconstruction of gas temperature profile is sen- 
sitive to the initial input of S x especially the [3 parameter, 
whether or not we can reliably derive the temperature 
profile depends critically on the goodness of the single (3 
model fit to the X-ray surface brightness profile. We thus 
restrict ourselves to the X-ray flux-limited sample of 45 


Yan-Jie Xue and Xiang-Ping Wu: Temperature Profiles of Clusters 

Table 1. 


3r Sample 









To (kcV) 











r c (Mpc) 




«c0 (10~ 3 

cm" 3 ) 









17.80(12.30) + 

14.66(24.80) + 

13.15(13.21) + 





r a (Mpc) 

5.77(2.59) + 

4.82(5.99) + 

3.43(2.03) + 

*In units of 10 erg s cm arcmin for energy band 0.5-2.0 keV; 

The result for an isothermal gas distribution. 

clusters published recently by Mohr, Mathiesen & Evrard 
(1999), in which there are sufficiently large data points to 
set robust constraints on the (3 model fit. The inclusion 
of a cluster is based on the following two criteria: (l)Thc 
X-ray surface brightness profile can be well fitted by a 
single j3 model with 0.8 < xt — 1-25; (2)The maximum 
extension (r m ) of the X-ray observed surface brightness 
profile should be large enough to guarantee the validity 
of the (3 model at the outermost regions of clusters. Here 
we set r m > 1.5 Mpc. Unfortunately, it turns out that 
there are only three clusters which meet our criteria (Ta- 
ble 1): A119, A2255 and A2256. In fact, our first criterion 
implies that the effect of cooling flows in the central re- 
gions of clusters should be negligibly small. This explains 
the fact that the three selected clusters all have large core 
radii. Note that the presence of cooling flows may lead 
to the failure of a single f3 model fit to the X-ray surface 
brightness profiles. In other words, our method cannot be 
applied to the clusters with strong cooling flows. While the 
X-ray imaging data of the clusters can be somewhat accu- 
rately acquired, the present X-ray spectral measurements 
have yielded the emission-weighted temperatures rather 
than the central values To appearing in the «o parameter. 
Therefore, we have to use the emission-weighted tempera- 
ture as a first approximation of To. Alternatively, we adopt 
the universal baryon fraction /b,BBN = 1/6 to reconcile 
our cosmological model of f^M = 0.3 (for fib = 0.05). 

Using the available X-ray data of the three clusters 
from Mohr et al. (1999), we have performed numerical 
searches for the solutions of eqs.(8) and (9) by iterations 
until the boundary conditions eqs.(10) and (11) are sat- 
isfied. The resulting parameters an,, S c and r s are sum- 
marized in Table 1, together with a comparison with the 
corresponding values for an isothermal gas distribution 
estimated in previous work (Wu & Xue 2000). Most im- 
portantly, such a procedure enables us to completely fix 
the radial profiles of gas density, temperature and baryon 
fraction for the three clusters. Here we have no intention 
to illustrate the radial variations of n e (r) and fb(f), which 

r/r ■ 

' vir 

Fig. 1. A comparison of the derived radial temperature 
profiles of three clusters (A119, A2255 and A2256) with 
the results of 1 1 clusters observed with BeppoS AX (Irwin 
& Bregman 2000). The observed data are normalized by 
the mean temperature for each cluster, while the derived 
temperature curves are scaled by 1.32To for comparison. 
The horizontal axis is in units of the virial radius. 

essentially follow the theoretical expectations (Wu & Chi- 
ueh 2000). Rather, we display in Fig.l the radial profiles 
of the emission-weighted temperatures for the three clus- 
ters constructed from our algorithm. Surprisingly, none of 
the temperature profiles of these three clusters are con- 
sistent with the conventional speculations, and a visual 
examination of Fig.l reveals that they are neither char- 
acterized by isothermality nor represented simply by the 
polytropic equation of state. Nevertheless, these tempera- 
ture profiles indeed demonstrate a similar radial variation, 
reflecting probably the underlying structural regularity. 
Basically, the radial variation of the gas temperature re- 
sembles a distorted 'S' in shape: There exist two turnover 
points roughly at 0.1r v j r and 0.4r v j r , respectively, where 
dT/dr = 0, which separate the temperature curve T{r) 
into three parts - a decreasing T(r) with radius inside the 
cluster core of ~ 0.1r v ; r , following a slightly increasing 
T(r) until ~ 0.4r v i r , and finally a moderately decreasing 
T(r) out to the virial radius. Overall, the absolute values 
of the gas temperature do not demonstrate a dramatic 
change within clusters. 

The azimuthally-averaged radial temperature profiles 
of 11 clusters derived by Irwin & Bregman (2000) from 
an analysis of the BeppoSAX data are superimposed on 
Fig.l. It appears that our derived temperature profiles 
are in good agreement with their observed ones over en- 
tire radius range. In fact, the significant temperature dis- 
crepancy raised in different X-ray spectral measurements 
occurs in the inner parts of clusters. In the outer regions, 

Yan-Jie Xue and Xiang-Ping Wu: Temperature Profiles of Clusters 

it seems that many observations have provided a moder- 
ately decreasing temperature profile, which is essentially 
consistent with our theoretical predictions. Alternatively, 
our result is also compatible with the gas temperature dis- 
tribution at large radii revealed by numerical simulations 
that demonstrate a temperature decline of ~ 30% of the 
central value at the virial radius (Frcnk ct al. 1999). 

4. Discussion and conclusions 

In the absence of the detailed information on the radial 
temperature profiles of clusters from X-ray spectroscopic 
measurements, we have made an attempt to derive the 
gas temperature profiles by combining the X-ray surface 
brightness measurements and the NFW profile as the un- 
derlying dark matter distribution of clusters. This has be- 
come possible when the intracluster gas is required to sat- 
isfy the hydrostatic equilibrium and the volume-averaged 
baryon fraction within the viral radius is required to 
asymptotically match the universal value. Consequently, 
we have obtained semi-analytically the temperature pro- 
files of three clusters selected carefully from the ROSAT 
observed cluster sample. These derived temperature pro- 
files are consistent with the new observations of 11 Bep- 
poSAX clusters (Irwin & Bregman 2000) and other mea- 
surements made at large cluster radii (e.g. Markevitch et 
al. 1998) as well as the result given by numerical simula- 
tions (e.g. Frenk et al. 1999). 

Regardless of the small sample, the three clusters ex- 
hibit a temperature profile similar in shape when the 
length scales are normalized to their virial radii, perhaps 
indicative of the underlying structural regularity. The 
present study provides a helpful clue to resolving the tem- 
perature profile discrepancy: It is very likely that the lack 
of the high-quality data of the spatially resolved spectral 
observations would yield an emission-weighted tempera- 
ture roughly close to isothcrmality within ~ 80% of the 
virial radius, which does not exclude the possibility that a 
slightly increasing temperature profile may be marginally 
detectable in the range of 0.1r v j r < r < 0.4r v j r . This ex- 
plains the recent observations of Irwin & Bregman (2000) 
and other studies (e.g. Kikuchi et al. 1999; White 2000; 
etc.). However, our model does not predict the flat tem- 
perature profile toward the inner regions of clusters as re- 
ported particularly by Markevitch et al. (1998), although a 
moderately decreasing temperature profile will ultimately 
take place in the outer clusters (r > 0.4r v ; r ). 

A conclusive test for the universality of our derived 
temperature profiles can be provided by future X-ray 
spectroscopic measurements. Indeed, it will be useful to 
apply the present method to other X-ray clusters with 
good X-ray surface brightness profiles measured to large 
radii and high-quality data of the spatially-resolved spec- 
tral observations at least within the central regions. This 
may allow us to further justify our model and include 
the measurement uncertainties which have been neglected 

in the present study. The inconsistency of the predicted 
temperature profiles with the X-ray spectroscopic results 
will challenge the prevailing models of structure forma- 
tions as well as the conventional scenario of cluster dy- 
namics such as the hydrostatic equilibrium. Finally, we 
should point out that our proposed method to obtain 
the temperature profiles of clusters can be significantly 
contaminated by nongravitational heating processes espe- 
cially from the supernova-driven protogalactic winds. Re- 
call that the asymptotic tendency of the derived temper- 
ature profiles at large radii depends sensitively on the (3 
parameter, while the energy injection of supernovae and 
active galaxies into the intracluster gas will result in a 
shallower X-ray surface brightness distribution (David et 
al. 1990; Ponman, Cannon & Navarro 1999; Llyod-Davics, 
Ponman & Cannon 2000). Without correction to this ef- 
fect the theoretically predicted temperature profiles may 
rise too rapidly at large radii. Note that at large radii the 
NFW mass profile diverges logarithmically with r, which 
differs significantly from the variation of the gas mass pro- 
file (roughly M gas oc r) expected from the assumption of 
isothcrmality. For a cluster with smaller f3, r c and r s , an 
increasing temperature profile near virial radius is thus re- 
quired to maintain the universality of the cluster baryon 
fraction. Therefore, a robust, theoretical determination of 
the temperature profiles of clusters should also allow non- 
gravitational heating processes to be included. 

Acknowledgements. We gratefully acknowledge Tzihong Chi- 
ueh for useful discussion and comments, and an anonymous ref- 
eree for valuable suggestions. This work was supported by the 
National Science Foundation of China, under Grant 19725311. 


Cavaliere A., Fusco-Femiano R., 1976, A&A, 49, 137 

Cowie L. L., Henriksen M., Mushotzky R. F., 1987, ApJ, 317, 

David L. P., Arnaud K. A., Forman W., Jones C, 1990, ApJ, 

356, 32 
Frenk C. S., et al., 1999, ApJ, 525, 554 
Irwin J. A., Bregman J. N., 2000, ApJ, in press 
Irwin J. A., Bregman J. N., Evrard A. E., 1999, ApJ, 519, 518 
Kikuchi K., et al., 1999, PASJ, 51, 301 
Lloyd-Davics E. J., Ponman T. J., Cannon D. B., 2000, MN- 

RAS, in press 
Markevitch M., Vikhlinin A., Forman W. R., Sarazin C. L., 

1998, ApJ, 527, 545 
Mohr J. J., Mathiesen B., Evrard A. E., 1999, ApJ, 517, 627 
Navarro J. F., Frenk C. S., White S. D. M., 1995, MNRAS, 

275, 720 (NFW) 
Neumann D. M., Arnaud M., 1999, A&A, 348, 711 
Ponman T. J., Cannon D. B., Navarro J. F., 1999, Nature, 397, 

White D. A., 2000, MNRAS, 312, 663 
Wu X.-R, Chiueh T., 2000, ApJ, in press 
Wu X.-P, Xue, Y.-J., 2000, ApJ, 529, L5